Traditional Picture Archiving Communications Systems (PACSs) have evolved primarily to serve the needs of the clinical settings in which they are predominantly used. Other settings such as medical imaging research, teaching, or industry, when faced with the need to store and catalogue DICOM image data, have typically (a) stored images as DICOM files on a typical computer file system (e.g. Windows™ NTFS™), using directory structures as a means of organizing the data, or (b) attempted to adapt traditional PACS solutions, working within, or around, their limitations, or (c) built custom “homegrown” systems, typically supplementing elements of (a) and/or (b) with additional external database(s). Each of these approaches have drawbacks and limitations.
The standard file system storage approach has the advantage of flexibility, in that users can organize their data in directory structures of arbitrary design. However, there are various drawbacks to this approach. For example, the user can interact with the DICOM images only as opaque files. Users cannot “see” or interact with the hierarchical study-series-image structure of the data, or other meta-data encoded in the DICOM files, without having to open the files in a DICOM application. Also, the user cannot search for images of interest, except by browsing the directory structure or searching based on file name. Because the files are opaque to the file system, DICOM metadata is not indexed and hence cannot be used as a basis for search. Using this approach, it is difficult to centralize the system for shared access. Network file systems support shared access, but provide no means of coordinating file access to prevent or resolve update conflicts and maintain data integrity. Finally, there is no support for DICOM communication protocols, which limits the ability to integrate with other DICOM-enabled devices and applications.
Adapting a traditional PACS has the advantage that, unlike the file system approach, PACSs are DICOM aware. They support DICOM communication protocols, typically allowing some degree of interaction with the hierarchical study-series-image structure of the data and/or other metadata encoded in the DICOM files, and supporting some basic search capabilities with respect to this metadata. Use of such a system also solves the problem of centralization for shared access. However, such an approach has several deficiencies.
PACSs limit the ability to organize and catalogue the data in arbitrary ways. The system organizes data to meet clinical needs, i.e., according to Patient and Study and Image identifiers. A corollary to this is that a given image can typically be stored exactly once in the system, because its identity is determined by its Image and Study identifiers. In non-clinical usage, this restriction is unnecessarily limiting, as there may be a need to store different copies of objects that share the same identifier.
Search capabilities offered by PACSs are typically limited to a very small subset of the DICOM metadata, which is selected for clinical relevance. A typical PACS might support searching by Patient ID, Patient Name, Study ID, Study Date, Study Description and Modality. The vast majority of the DICOM metadata is not indexed and hence cannot be used as a basis for search.
There is no ability to store native non-DICOM data in the system alongside DICOM data. With the file system approach, it is easy to add arbitrary (non-DICOM) files into a directory along with related image files, but storage of non-DICOM files isn't generally possible in a PACS, at least not without first encapsulating em into a DICOM container file.
Even PACSs that do support storage of encapsulated non-DICOM files do not typically support searches based on the content of these files. With the file system approach, typical operating systems (e.g. Windows™) tend to index the content of common file types (e.g. PDF), making it possible to search for files based on their content. In contrast, when a non-DICOM file is stored in a PACS (by encapsulating it into a DICOM container file), the file becomes opaque and its content is not searchable.
In such approaches, there is limited or no ability to “own” data. Because the PACS is designed for clinical use, the data typically resides in a single logical repository accessible to all users of the system. Ownership of individual pieces of data, either by individual users, or groups of users, is not possible.
DICOM communication may be the only way to import/export data to/from the PACS, which limits the ability to integrate it with applications that do not support DICOM communication.